Mingin Pingin (long)

Well mingin traceroute actually.

I've just changed ISP and been upgraded from 512k to 2M. When it's fast
it's great but at peak times I'm sure it's slower than my old 512k provider
ever was, speeds troughed out at 20kB/s last night and I ran a reverse
traceroute from my US NSP to get a feel where the problem was.

The results point to a problem between my ISP and me but I could do with
an experienced eye to suggest whether this might be:
1. Expected congestion
2. Hardware problems, theirs or mine
3. Underprovisioned exchange
4. Underprovisioned ISP
5. Really, really shit ISP
6. Other

The measured delay between the ISP & me was 175ms!!!!!! but note that
it's a reasonable 15ms outside peak times.

If it helps, I'm on a premium package but I'm not mentioning the ISP just
yet, they are sort-of on the case too. First trace is the crap one (20kB/s),
second is normal (off-peak) service, fixed font is best:

Poor:
Started on Tuesday August 16th, 2005 at 19:15:56 GMT
HOST LOSS RCVD SENT BEST AVG WORST
puregig.net 0% 10 10 0.56 6.80 36.77
Phoenix1.L3.net 0% 10 10 0.64 23.75 100.06
NewYork1.L3.net 0% 10 10 60.65 64.41 86.56
London1.Lvl3.net 0% 10 10 126.17 130.55 159.05
London1.Lvl3.net 0% 10 10 125.99 128.10 145.55
ge0.pte-gw2.xxx 0% 10 10 126.28 126.51 126.78
gi9.pte-ag1.xxx 0% 10 10 127.26 131.55 152.96
<mymachine> 10% 9 10 282.23 306.68 318.79

Recovered:
Started on Wednesday August 17th, 2005 at 01:42:10 GMT
HOST LOSS RCVD SENT BEST AVG WORST
puregig.net 0% 10 10 0.49 3.49 28.72
Phoenix1.L3.net 0% 10 10 0.64 2.69 15.03
NewYork1.L3.net 0% 10 10 60.65 61.43 67.23
London1.Lvl3.net 0% 10 10 126.03 126.13 126.28
London1.Lvl3.net 0% 10 10 126.00 130.38 168.01
ge0.pte-gw2.xxx 0% 10 10 126.32 130.28 152.93
gi9.pte-ag1.xxx 0% 10 10 127.07 136.48 168.01
<mymachine> 10% 9 10 149.97 151.75 161.13
Your max throughput is 175.61 KB/sec

--
fred

On Wed, 17 Aug 2005 15:05:33 GMT, fred <(E-Mail Removed)> wrote:


>Poor:
>Started on Tuesday August 16th, 2005 at 19:15:56 GMT
>HOST LOSS RCVD SENT BEST AVG WORST
>gi9.pte-ag1.xxx 0% 10 10 127.26 131.55 152.96
><mymachine> 10% 9 10 282.23 306.68 318.79
>
>Recovered:
>Started on Wednesday August 17th, 2005 at 01:42:10 GMT
>HOST LOSS RCVD SENT BEST AVG WORST
>gi9.pte-ag1.xxx 0% 10 10 127.07 136.48 168.01
><mymachine> 10% 9 10 149.97 151.75 161.13
>Your max throughput is 175.61 KB/sec

I've had a guess at your ISP based on router naming conventions.

Knowing the ISP helps to determine how your BB connection is
provisioned (i.e. IP Stream, Datastream or LLU), and if it's who I
think it is then my gut feeling would be a congestion problem that
lies within BT between your exchange and the Colossus network at peak
times, not a problem with your ISP (unless it's their own Centrals
that max out at peak times).

Jake

In article <(E-Mail Removed)>, Muxton
<(E-Mail Removed)> writes
>On Wed, 17 Aug 2005 15:05:33 GMT, fred <(E-Mail Removed)> wrote:
>
>
>>Poor:
>>Started on Tuesday August 16th, 2005 at 19:15:56 GMT
>>HOST LOSS RCVD SENT BEST AVG WORST
>>gi9.pte-ag1.xxx 0% 10 10 127.26 131.55 152.96
>><mymachine> 10% 9 10 282.23 306.68 318.79
>>
>>Recovered:
>>Started on Wednesday August 17th, 2005 at 01:42:10 GMT
>>HOST LOSS RCVD SENT BEST AVG WORST
>>gi9.pte-ag1.xxx 0% 10 10 127.07 136.48 168.01
>><mymachine> 10% 9 10 149.97 151.75 161.13
>>Your max throughput is 175.61 KB/sec
>
>I've had a guess at your ISP based on router naming conventions.
>
>Knowing the ISP helps to determine how your BB connection is
>provisioned (i.e. IP Stream, Datastream or LLU), and if it's who I
>think it is then my gut feeling would be a congestion problem that
>lies within BT between your exchange and the Colossus network at peak
>times, not a problem with your ISP (unless it's their own Centrals
>that max out at peak times).

I see, thanks, it's Plusnet (the 30:1 premier service), I was trying to avoid
colouring any opinions. I suspected it might just be a congestion issue but
thought that 175ms for the last hop was a bit of a joke. They've asked me
for BT speedtests so will be doing them tonight.
--
fred

"fred" <(E-Mail Removed)> wrote in message news:m$(E-Mail Removed)...
> In article <(E-Mail Removed)>, Muxton
> <(E-Mail Removed)> writes
>>On Wed, 17 Aug 2005 15:05:33 GMT, fred <(E-Mail Removed)> wrote:
>>
>>
>>>Poor:
>>>Started on Tuesday August 16th, 2005 at 19:15:56 GMT
>>>HOST LOSS RCVD SENT BEST AVG WORST
>>>gi9.pte-ag1.xxx 0% 10 10 127.26 131.55 152.96
>>><mymachine> 10% 9 10 282.23 306.68 318.79
>>>
>>>Recovered:
>>>Started on Wednesday August 17th, 2005 at 01:42:10 GMT
>>>HOST LOSS RCVD SENT BEST AVG WORST
>>>gi9.pte-ag1.xxx 0% 10 10 127.07 136.48 168.01
>>><mymachine> 10% 9 10 149.97 151.75 161.13
>>>Your max throughput is 175.61 KB/sec
>>
>>I've had a guess at your ISP based on router naming conventions.
>>
>>Knowing the ISP helps to determine how your BB connection is
>>provisioned (i.e. IP Stream, Datastream or LLU), and if it's who I
>>think it is then my gut feeling would be a congestion problem that
>>lies within BT between your exchange and the Colossus network at peak
>>times, not a problem with your ISP (unless it's their own Centrals
>>that max out at peak times).
>
> I see, thanks, it's Plusnet (the 30:1 premier service), I was trying to
> avoid
> colouring any opinions. I suspected it might just be a congestion issue
> but
> thought that 175ms for the last hop was a bit of a joke. They've asked me
> for BT speedtests so will be doing them tonight.
> --
> fred

My guess is that you have picked up another satellite connection on the
slower ping, because the cable connection is busy. This adds ~72Mm of free
space to your connection path, which assuming Einstein was relatively
correct will add about a quarter of a second latency to the ping. Might or
might not affect bandwidth.

In article <ddvu54$ic3$(E-Mail Removed)>, R.
Mark Clayton <(E-Mail Removed)> writes
>
>"fred" <(E-Mail Removed)> wrote in message news:m$(E-Mail Removed)...
>> In article <(E-Mail Removed)>, Muxton
>> <(E-Mail Removed)> writes
>>>On Wed, 17 Aug 2005 15:05:33 GMT, fred <(E-Mail Removed)> wrote:
>>>
>>>
>>>>Poor:
>>>>Started on Tuesday August 16th, 2005 at 19:15:56 GMT
>>>>HOST LOSS RCVD SENT BEST AVG WORST
>>>>gi9.pte-ag1.xxx 0% 10 10 127.26 131.55 152.96
>>>><mymachine> 10% 9 10 282.23 306.68 318.79
>>>>
>>>>Recovered:
>>>>Started on Wednesday August 17th, 2005 at 01:42:10 GMT
>>>>HOST LOSS RCVD SENT BEST AVG WORST
>>>>gi9.pte-ag1.xxx 0% 10 10 127.07 136.48 168.01
>>>><mymachine> 10% 9 10 149.97 151.75 161.13
>>>>Your max throughput is 175.61 KB/sec
>>>
>>>I've had a guess at your ISP based on router naming conventions.
>>>
>>>Knowing the ISP helps to determine how your BB connection is
>>>provisioned (i.e. IP Stream, Datastream or LLU), and if it's who I
>>>think it is then my gut feeling would be a congestion problem that
>>>lies within BT between your exchange and the Colossus network at peak
>>>times, not a problem with your ISP (unless it's their own Centrals
>>>that max out at peak times).
>>
>> I see, thanks, it's Plusnet (the 30:1 premier service), I was trying to
>> avoid
>> colouring any opinions. I suspected it might just be a congestion issue
>> but
>> thought that 175ms for the last hop was a bit of a joke. They've asked me
>> for BT speedtests so will be doing them tonight.
>> --
>> fred
>
>My guess is that you have picked up another satellite connection on the
>slower ping, because the cable connection is busy. This adds ~72Mm of free
>space to your connection path, which assuming Einstein was relatively
>correct will add about a quarter of a second latency to the ping. Might or
>might not affect bandwidth.

They're using satellite in the UK now :-?
The big hops from the original post are: 60ms for the big hop to New York,
60ms for the transatlantic hop then a huge 175ms just for the last hop from
my (UK) isp to my machine. This was the story from about 8-10pm last
night.
--
fred

On Wed, 17 Aug 2005 18:03:17 +0000 (UTC), "R. Mark Clayton"
<(E-Mail Removed)> wrote:

>
>"fred" <(E-Mail Removed)> wrote in message news:m$(E-Mail Removed)...
>> In article <(E-Mail Removed)>, Muxton
>> <(E-Mail Removed)> writes
>>>On Wed, 17 Aug 2005 15:05:33 GMT, fred <(E-Mail Removed)> wrote:
>>>
>>>
>>>>Poor:
>>>>Started on Tuesday August 16th, 2005 at 19:15:56 GMT
>>>>HOST LOSS RCVD SENT BEST AVG WORST
>>>>gi9.pte-ag1.xxx 0% 10 10 127.26 131.55 152.96
>>>><mymachine> 10% 9 10 282.23 306.68 318.79
>>>>
>>>>Recovered:
>>>>Started on Wednesday August 17th, 2005 at 01:42:10 GMT
>>>>HOST LOSS RCVD SENT BEST AVG WORST
>>>>gi9.pte-ag1.xxx 0% 10 10 127.07 136.48 168.01
>>>><mymachine> 10% 9 10 149.97 151.75 161.13
>>>>Your max throughput is 175.61 KB/sec
>>>
>>>I've had a guess at your ISP based on router naming conventions.
>>>
>>>Knowing the ISP helps to determine how your BB connection is
>>>provisioned (i.e. IP Stream, Datastream or LLU), and if it's who I
>>>think it is then my gut feeling would be a congestion problem that
>>>lies within BT between your exchange and the Colossus network at peak
>>>times, not a problem with your ISP (unless it's their own Centrals
>>>that max out at peak times).
>>
>> I see, thanks, it's Plusnet (the 30:1 premier service), I was trying to
>> avoid
>> colouring any opinions. I suspected it might just be a congestion issue
>> but
>> thought that 175ms for the last hop was a bit of a joke. They've asked me
>> for BT speedtests so will be doing them tonight.
>> --
>> fred
>
>My guess is that you have picked up another satellite connection on the
>slower ping, because the cable connection is busy. This adds ~72Mm of free
>space to your connection path, which assuming Einstein was relatively
>correct will add about a quarter of a second latency to the ping. Might or
>might not affect bandwidth.
>

It's the last hop you need to look at (which would actually be the
first hop from the OP's POV), all the other hops are irrelevant.

Jake

On Wed, 17 Aug 2005 17:35:23 GMT, fred <(E-Mail Removed)> wrote:


>>Knowing the ISP helps to determine how your BB connection is
>>provisioned (i.e. IP Stream, Datastream or LLU), and if it's who I
>>think it is then my gut feeling would be a congestion problem that
>>lies within BT between your exchange and the Colossus network at peak
>>times, not a problem with your ISP (unless it's their own Centrals
>>that max out at peak times).
>
>I see, thanks, it's Plusnet

I got that right then.

>(the 30:1 premier service)

Don't know how they provision that, given that AFAIK they use IP
Stream. Is that IP Stream 20:1 with a bit of Plusnet contention
thrown in to create a hybrid? Weird.

>I was trying to avoid
>colouring any opinions. I suspected it might just be a congestion issue but
>thought that 175ms for the last hop was a bit of a joke. They've asked me
>for BT speedtests so will be doing them tonight.

Be sure to post the result here. I'd still suspect local issues
rather than Plusnet issues, but now that you mention the 30:1 thing it
seems PN might be doing odd things with the traffic.

Jake

In article <(E-Mail Removed)>, Muxton
<(E-Mail Removed)> writes
>On Wed, 17 Aug 2005 17:35:23 GMT, fred <(E-Mail Removed)> wrote:
>
>
>>(the 30:1 premier service)
>
>Don't know how they provision that, given that AFAIK they use IP
>Stream. Is that IP Stream 20:1 with a bit of Plusnet contention
>thrown in to create a hybrid? Weird.
>
>>I was trying to avoid
>>colouring any opinions. I suspected it might just be a congestion issue but
>>thought that 175ms for the last hop was a bit of a joke. They've asked me
>>for BT speedtests so will be doing them tonight.
>
>Be sure to post the result here. I'd still suspect local issues
>rather than Plusnet issues, but now that you mention the 30:1 thing it
>seems PN might be doing odd things with the traffic.

Throughput is currently disappointing (56kB/448kb/s from 2Mb) rather than
diabolical so I will save my indignance for when it is really pants. Monday
and Tuesday evenings seem to be super-peaks where my end of their
system seems to fall apart, I'll do some tests then but in the mean time
real life goes on . . .

As it will be a week or so on I will follow-up in a new thread, keep a look
out for mingin followup/update.
--
fred

On Wed, 17 Aug 2005 22:25:36 GMT, fred <(E-Mail Removed)> wrote:


>As it will be a week or so on I will follow-up in a new thread, keep a look
>out for mingin followup/update.

Will do, you have a week to think of another amusing rhyming subject
line ;-)

Jake

On Wed, 17 Aug 2005 21:37:01 +0100, Muxton
<(E-Mail Removed)> wrote:

>Don't know how they provision that, given that AFAIK they use IP
>Stream. Is that IP Stream 20:1 with a bit of Plusnet contention
>thrown in to create a hybrid? Weird.

its IPStream "up to 50:1" from BT with 30:1 contention "on the Plusnet
network" in other words they allocate a 2M user 67 k on their BT
Central links when planning capacity.

Phil
--
Tiscali - dialup speeds at Broadband prices, see
http://bbs.adslguide.org.uk/postlist...&Board=tiscali

AOL - the unlimited ISP of choice for heavy downloaders.